What Is Reliabilism?

Direct Answer

Reliabilism is a theory of epistemic justification that locates the source of justification not in the believer's evidence or mental states but in the reliability of the process that produced the belief. The core idea is straightforward: a belief is justified if it was formed by a cognitive process that tends to produce true beliefs. If your visual perception is working properly and you see a tree in front of you, your belief that there is a tree is justified — not because you have evidence that there is a tree, but because the process of visual perception is generally reliable, producing true beliefs far more often than false ones.

This makes reliabilism an externalist theory of justification. Internalist theories like evidentialism hold that the justifying factors must be accessible to the believer — you must be able to know, by reflection, what makes your belief justified. Reliabilism denies this. The reliability of your perceptual system is a fact about the world and your cognitive machinery, not a fact you need to be aware of. A young child who sees a tree has a justified belief that there is a tree, even though the child has no beliefs about the reliability of perception. The justifying factor — the reliability of the process — is external to the child's perspective.

The contrast with internalism is sharp. Suppose two people have identical internal mental states — the same experiences, the same beliefs, the same evidence — but one lives in a normal environment where perception is reliable and the other lives in an environment where a deceiving demon makes perception unreliable. The internalist must say that both have the same epistemic status, since their internal states are the same. The reliabilist says that the first person's beliefs are justified (because produced by a reliable process) while the second person's beliefs are not (because produced by an unreliable process). This difference is the crux of the internalism-externalism debate in epistemology.

Alvin Goldman, the leading figure in the development of reliabilism, initially formulated the view in his 1979 paper "What Is Justified Belief?" and later refined it in Epistemology and Cognition (1986). Goldman's version is called "process reliabilism" because it identifies the justifying factor as the reliability of the belief-forming process. Other versions of reliabilism have been developed — including "virtue reliabilism," which ties justification to the exercise of intellectual virtues, and "proper functionalism," which ties justification to the proper functioning of cognitive faculties — but process reliabilism remains the paradigmatic version.

Historical Context

The Stanford Encyclopedia of Philosophy notes that reliabilism emerged in the 1960s and 1970s as part of a broader shift in epistemology from internalist to externalist approaches. The immediate stimulus was the Gettier problem — Edmund Gettier's 1963 demonstration that justified true belief is not sufficient for knowledge. Gettier presented cases in which a person has a belief that is both justified and true, but in which the truth of the belief is due to luck rather than to the justification. These cases showed that the traditional analysis of knowledge as justified true belief was incomplete, and they prompted philosophers to look for a fourth condition — something that would rule out Gettier cases.

Goldman was among the first to propose that the missing condition was a causal connection between the fact that makes the belief true and the belief itself. In his 1967 paper "A Causal Theory of Knowing," Goldman argued that knowledge requires the belief to be caused by the fact that makes it true. This causal theory was a precursor to reliabilism: it located the epistemically relevant factor in the causal connection between the world and the belief, not in the believer's internal evidence. The causal theory had limitations — it struggled with beliefs about the future, generalizations, and mathematical truths — but it opened the door to externalist approaches.

Process reliabilism was Goldman's more general and more successful proposal. Instead of requiring a direct causal connection between fact and belief, it required that the belief be produced by a process that is generally reliable — that tends to produce true beliefs across a range of cases. This was more flexible than the causal theory: it could handle beliefs about the future (the process of inductive reasoning can be reliable even without direct causal connections) and beliefs based on testimony (the process of accepting testimony from reliable sources can be reliable).

The emergence of reliabilism coincided with developments in cognitive science and the philosophy of mind that were friendly to externalist approaches. The cognitive revolution of the 1950s and 1960s had made it natural to think of belief formation as a process — something done by cognitive mechanisms operating according to certain principles. If belief formation is a process, then it makes sense to ask whether the process is reliable, and to tie epistemic justification to the answer. Reliabilism fit the spirit of the age: it was naturalistic, scientifically informed, and focused on the mechanisms of cognition rather than the introspective evidence of the believer.

Philosophical Perspectives

Process reliabilism faces several significant objections that have shaped the debate. The first is the generality problem. To evaluate whether a belief is justified, the reliabilist must identify the process that produced it and determine whether that process is reliable. But processes can be described at different levels of generality. The process that produced your belief that there is a tree might be described as "visual perception" (which is generally reliable), or as "visual perception in good lighting conditions" (which is very reliable), or as "visual perception while tired on a foggy morning" (which might be less reliable). The reliability of the process varies depending on how it is described, and the reliabilist needs a principled way to select the right level of generality. Without it, the theory is indeterminate: the same belief can be both justified and unjustified depending on how we describe the process.

Goldman has acknowledged the generality problem as the most serious challenge to process reliabilism, and he has proposed various solutions, including the idea that processes should be typed by their "psychological" features — the features that are psychologically real to the agent, not merely features imposed by the theorist's description. But critics argue that no fully satisfactory solution has been found, and the generality problem remains a central topic in the reliabilism literature.

A second objection is the new evil demon problem. Imagine a person whose cognitive processes are qualitatively identical to yours — they have the same experiences, form the same beliefs, reason in the same ways — but who lives in a world controlled by an evil demon who makes their perceptions systematically misleading. Their perceptual beliefs are mostly false, because the demon creates illusions. By the reliabilist's standard, this person's beliefs are not justified, because the process producing them is unreliable. But this seems counterintuitive: from the inside, the demon-victim's epistemic situation is indistinguishable from yours, and it seems harsh to say that they are unjustified in their beliefs. The internalist can handle this case easily — both you and the demon-victim have the same evidence, so you have the same justification — but the reliabilist seems committed to saying that the demon-victim is epistemically deficient despite having done everything right from their perspective.

Reliabilists have responded in various ways. Some bite the bullet and accept that the demon-victim's beliefs are not justified, arguing that this is a feature rather than a bug of the theory: justification should track the real reliability of the process, not the apparent reliability from the inside. Others modify the theory, perhaps by saying that justification depends on reliability in normal environments, or by introducing a "normal world" clause. The Stanford Encyclopedia of Philosophy notes that the evil demon problem remains a live issue, and it has led some reliabilists to incorporate internalist elements into their view.

A third objection concerns the relationship between reliability and reflective access. If justification depends on reliability, and reliability is an external fact about the process, then it seems that a person could be justified in a belief without knowing that they are justified — without having any reflective access to the justifying factor. This raises the question of whether reliabilist justification can play the role that justification is supposed to play in rational inquiry. If I cannot tell, by reflection, whether my belief is justified, how can I use justification as a guide to which beliefs to hold and which to abandon? The internalist argues that justification must be accessible to serve this guiding role; the reliabilist responds that the guiding role is played by the belief-forming process itself, not by the believer's awareness of its reliability.

Despite these challenges, reliabilism has been enormously influential. It offered a fresh perspective on epistemic justification that broke free from the internalist assumptions that had dominated epistemology since Descartes. It connected epistemology to cognitive science by treating belief formation as a natural process amenable to empirical study. And it provided a natural solution to the Gettier problem: if justification requires reliable processes, then Gettier cases — where true belief is produced by lucky but unreliable processes — are ruled out. A belief that is both true and produced by a reliable process is knowledge; a belief that is true but produced by luck is not.

Modern Reflection

Reliabilism has evolved significantly since Goldman's original formulation. One important development is the emergence of virtue epistemology, which shares with reliabilism the idea that justification depends on the quality of the belief-forming process but reframes the quality in terms of intellectual virtues rather than reliability. Ernest Sosa's virtue perspectivism holds that a belief is justified if it is produced by a virtuous cognitive faculty — one that is accurate, adroit, and apt. Sosa's framework incorporates both externalist elements (the reliability of the faculty) and internalist elements (the agent's reflective perspective on their own cognition), creating a hybrid that addresses some of the objections to pure process reliabilism.

Alvin Plantinga's proper functionalism is another externalist development that shares DNA with reliabilism. Plantinga argues that a belief is warranted — his term for the epistemic property that turns true belief into knowledge — if and only if it is produced by cognitive faculties functioning properly in the environment for which they were designed. This is externalist in the same way that reliabilism is: the justifying factor (proper function) is not accessible to the believer. But it replaces the statistical notion of reliability with the teleological notion of proper function, which some philosophers find more intuitive.

Reliabilism has also influenced debates in applied epistemology. In the epistemology of education, reliabilist ideas have been used to argue that the goal of education should be the cultivation of reliable belief-forming processes — not just the transmission of information but the training of cognitive faculties to produce true beliefs. In the epistemology of testimony, reliabilism provides a framework for thinking about when testimony is a reliable source of belief: testimony from experts, from institutions with good track records, and from multiple independent sources is more reliable than testimony from a single, unverified source.

The rise of artificial intelligence and machine learning has given reliabilism new relevance. When a machine learning model makes a prediction, the question of whether the prediction is "justified" is, in reliabilist terms, a question about the reliability of the model's inference process. A model that has been trained on representative data, validated on held-out data, and tested under realistic conditions is more reliable than one that has not. The reliabilist framework provides a natural way to think about the epistemic status of AI-generated beliefs, though it also raises new versions of old problems: the generality problem reappears as the question of how to specify the relevant domain of the model's operation, and the evil demon problem reappears as the question of what happens when the model encounters distribution shift.

Related Thinkers

Aristotle discussed the role of cognitive faculties in producing knowledge, though his framework is not reliabilist in the modern sense. Alvin Goldman is the founding figure of process reliabilism, having developed the view from the 1970s onward. His 1979 paper "What Is Justified Belief?" and his 1986 book Epistemology and Cognition remain the foundational texts. Ernest Sosa's virtue epistemology extends and modifies reliabilist ideas by grounding justification in intellectual virtue rather than process reliability. Alvin Plantinga's proper functionalism offers a different externalist alternative, tying warrant to the proper functioning of cognitive faculties. On the internalist side, Richard Feldman and Earl Conee's evidentialism has been the main foil for reliabilist arguments.

Related Quotes

Goldman's original formulation of the reliabilist principle — that "a belief is justified if and only if the process that produces it is reliable" — remains the canonical statement of the view. The Stanford Encyclopedia of Philosophy notes that while the precise formulation has been debated and refined, the core insight — that the epistemic status of a belief depends on the quality of the process that produces it — has proven remarkably durable and continues to shape contemporary epistemology.

Further Learning

1. Stanford Encyclopedia of Philosophy, Reliabilism.
2. Alvin Goldman, "What Is Justified Belief?" in Justification and Knowledge (Reidel, 1979).
3. Alvin Goldman, Epistemology and Cognition (Yale University Press, 1986).
4. Ernest Sosa, A Virtue Epistemology (Oxford University Press, 2007).
5. Alvin Plantinga, Warrant and Proper Function (Oxford University Press, 1993).
6. Marshall Swain, Reason and Knowledge (Cornell University Press, 1981).

To explore related concepts, consider reading about epistemology, evidentialism, and the nature of knowledge.